CyberTraining: Implementation: Small: Developing a Best Practices Training Program in Cyberinfrastructure-Enabled Machine Learning Research

网络培训：实施：小型：制定网络基础设施支持的机器学习研究最佳实践培训计划

• 批准号: 2017767
• 负责人: Mary Thomas
• 金额: $ 50万
• 依托单位: University of California-San Diego
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-09-01 至 2024-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2017767&HistoricalAwards=false
• 关键词: CyberTraining; Implementation; Small; Developing; Best

This project is targeted towards the NSF research workforce who need to develop machine learning applications that will run on the national cyberinfrastructure (CI). At the heart of this project is the CI-enabled Machine Learning (CIML) training system and repository that will support the development of cyberliteracy in the ML space. Unlike much of current ML-related training material found easily online, CIML material will be centered on science and engineering applications that make use of CI-enabled ML techniques and will be used to train a research workforce that is capable of understanding the challenges of working with CI, new HPC architectures, software, and applications. The user community for CIML training material will include students (undergraduate, graduate), postdocs, PIs, researchers, educators, and HPC trainers, each with their own diverse backgrounds and application requirements. The project will support national educational goals by ensuring that the CI modules run on advanced CI tools and resources, and that core literacy and discipline appropriate skills in advanced CI will be integrated into curricula and instructional material. CIML will support national security concerns by facilitating a workforce capable of developing ML applications in scientific domains such as climate and weather, the biosciences, physics, and chemistry. As a result of outreach and extension of the training efforts, this program will impact thousands of users and help develop the next generation of the CI research workforce. CIML training material will be available online, so the project has a huge potential to reach beyond the NSF cyber workforce to impact other communities including hospital and medical treatment systems, transportation and electrical monitoring systems, stock market monitoring systems, and disaster response systems. The Cyberinfrastructure-enabled Machine Learning (CIML) training system and repository will use a “best practices” approach to develop a unique program targeted towards the research workforce who use machine learning (ML) and big data analytics methods for their domain specific applications or instructional material on large-scale cyberinfrastructure. The project will apply methods of Cyber Literacy and HPC Competencies to define a set of core ML and domain specific literacy areas as a function of the dimensions of learning ranging from a technological focus to a problem-solving focus or a focus on ML or computational science. Sources for the CIML system will be drawn from the work of HPC training, existing HPC researchers and users, collaborators, as well as new code and methods. The materials developed will be available via the CIML repository, which includes a web site, documentation, GitHub repositories for code, data, and related materials. CIMIL will become a useful tool for 2 communities: users who want to understand what technologies and skills they need to master in order to run a particular ML application, what systems to use, and suggested software libraries; and trainers who need to know what topics to teach. The outcome of these efforts will result in a community of machine learning and data analytics CI Users (CIU) and Contributors (CIC) who actively contribute to the training material repository and incorporate the materials into their projects and courses. As a result of these efforts, the CIML program will extend the scope of the ongoing education and training across the research workforce by developing cyberinfrastructure-based materials that will utilize and contribute to training material developed for XSEDE training, higher education, and other programs, and will impact thousands of existing and new users, including students (undergrads/grads), postdocs, PIs, researchers, and educators, each with their own diverse backgrounds and application requirements. CIML training material will be available online, so the project has a huge potential to reach beyond the cyber workforce and to impact many communities, including hospital and medical treatment systems, transportation and electrical monitoring systems, stock and market monitoring systems, and disaster response systems.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

该项目针对的是NSF研究人员，他需要开发将在国家网络基础设施（CI）上运行的机器学习应用程序。该项目的核心是支持CI的机器学习（CIML）培训系统和存储库，该系统将支持ML空间中的网络文学发展。与当前与ML相关的许多培训材料不同，CIML材料将集中在使用CI-NABLED ML技术的科学和工程应用上，并将用于培训能够理解与CI，新的HPC体系结构，软件和应用程序所面临的挑战的研究人员。 CIML培训材料的用户社区将包括学生（本科生，研究生），博士后，PIS，研究人员，教育工作者和HPC培训师，每个培训师都有自己的潜水员背景和应用程序要求。该项目将通过确保CI模块运行高级CI工具和资源来支持国家教育目标，并将高级CI中的核心素养和纪律技能集成到课程和教学材料中。 CIML将通过支持能够在诸如气候和天气，生物科学，物理和化学等科学领域中开发ML应用的劳动力来支持国家安全问题。由于培训工作的推广和扩展，该计划将影响数千名用户，并帮助发展下一代CI研究人员。 CIML培训材料将在线提供，因此该项目具有巨大的潜力，可以超越NSF网络劳动力，以影响其他社区，包括医院和医疗系统，运输和电气监测系统，股票市场监测系统以及灾难响应系统。网络基础设施支持机器学习（CIML）培训系统和存储库将使用“最佳”实践”方法来开发针对使用机器学习（ML）的研究人员（ML）和大数据分析方法的独特计划，并针对其领域的特定应用程序或教学材料在大规模cyberfrasture的特定方法上应用cyberinfrasture的特定方法和HPC的范围。从技术重点到解决问题的重点或CIML系统来源的素养领域，将从HPC培训的工作，现有的HPC研究人员和用户，协作者，合作者以及新的代码和方法中，包括新的代码和材料，包括新的代码和方法，包括新代码和方法，包括新的Repos，包括新的repos，将包括新的REPOS，包括新的Repos，相关的材料。以及需要知道要教的主题的培训师。这些努力的结果将导致机器学习和数据分析CI用户（CIU）和贡献者（CIC）积极贡献培训材料存储库，并将材料纳入其项目和课程。 As a result of these efforts, the CIML program will extend the scope of the ongoing education and training across the research workforce by developing cyberinfrastructure-based materials that will utilize and contribute to training material developed for XSEDE training, higher education, and other programs, and will impact thousands of existing and new users, including students (undergrads/grads), postdocs, PIs, researchers, and educators, each with their own divers backgrounds and application requirements. CIML培训材料将在线提供，因此该项目具有超越网络劳动力的巨大潜力，并影响了许多社区，包括医院和医疗系统，运输和电气监测系统，股票和市场监控系统以及灾难响应系统。该奖项反映了NSF的法规任务，并认为通过基金会的知识优点和广泛的criperia criperia criperia criperia criperia the Insportaution supportiation the Priews criptuation。

项目成果

Experiences in Building a User Portal for Expanse Supercomputer

Expanse超级计算机用户门户构建经验

• DOI: 10.1145/3437359.3465590
• 发表时间: 2021
• 期刊: Practice & Experience in Advanced Research Computing (PEARC
• 作者: Sakai, Scott;Mishin, Dmitry;Sivagnanam, Subhashini;Tatineni, Mahidhar;Kandes, Martin;Thomas, Mary;Irving, Christopher;Strande, Shawn;Norman, Michael
• 通讯作者: Norman, Michael

Critique of: “A Parallel Framework for Constraint-Based Bayesian Network Learning via Markov Blanket Discovery” by SCC Team From UC San Diego

加州大学圣地亚哥分校 SCC 团队对“通过马尔可夫毯子发现进行基于约束的贝叶斯网络学习的并行框架”的评论

• DOI: 10.1109/tpds.2022.3217284
• 发表时间: 2023
• 期刊: IEEE Transactions on Parallel and Distributed Systems
• 影响因子: 5.3
• 作者: Gupta, Arunav;Ge, John;Li, John;Kong, Zihao;He, Kaiwen;Mikhailov, Matthew;Chin, Bryan;Li, Xiaochen;Apodaca, Max;Rodriguez, Paul
• 通讯作者: Rodriguez, Paul

Expanse : Computing without Boundaries

Expanse™：无边界计算

• 发表时间: 2021
• 期刊: Practice & Experience in Advanced Research Computing (PEARC
• 作者: Strande, Shawn;Altintas, Ilkay;Cai, Haisong;Cooper, Trevor;Irving, Christopher;Kandes, Marty;Majumdar, Amitava;Mishin, Dmitry;Perez, Ismael;Pfeiffer, Wayne
• 通讯作者: Pfeiffer, Wayne